Recycled Materials and Byproducts in Highway Applications—Manufacturing and Construction Byproducts, Volume 8 (2013)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

66 Post-consumer glass cullet can usually be separated by color: amber, green, and clear. The Northeast Resource Recovery Association identifies suitable sources of recycled crushed glass as glass or ceramic bottles, glass jars, ceramic table- ware and cookware, vases, ceramic flowerpots, plate glass, mirror glass, and residential incandescent light bulbs. Waste glass from material recovery facilities can be identified by several names: cullet, recycled glass, soda lime glass, crushed glass, or processed glass aggregate. This byproduct is recov- ered from glass containers and from breakages and inferior products made during glass manufacturing. The majority of post-consumer containers can be sorted into three categories based on color, which is defined by the chemical composition needed to produce the color: • Flint glass: colorless glass food, beverage, beer, liquor, and wine bottles. • Amber glass: brown beer and liquor bottles. • Green glass: green wine and beer bottles. Another source of waste glass is from the glass manufactur- ing process, which includes such materials as broken, obsolete, and/or off specification glass from the manufacturing of plate, window, and analytical glassware (Wartman et al. 2004). Glass from automobiles, lead crystal, TV monitors, lighting fixtures, and electronics applications are excluded owing to their com- position and coatings. Glass cullet can be provided by the material recovery facil- ities as unwashed larger broken glass particles, unwashed but crushed glass cullet, and as-washed glass cullet. Washing the byproduct removes most of the contaminates such as paper, plastics, and metals that would be considered contaminates in most highway applications. Physical and chemical ProPerties One example of the physical properties of glass cullet, not crushed, used by the Ramsey County Public Works Depart- ment in Minnesota is shown in Table 85 (Krivit 1999). The American Geophysical Institute (AGI) visual evaluation is a subjective method of defining the percent of contaminates in the cullet. In the case of unwashed cullet, the percent of contaminates could be well above the recommended limits of 5% (TFHRC 2010). The contaminates influenced a number of the other physical properties such as the moisture content, specific gravity, permeability, and biological and chemical content. Wartman et al. (2004) reported unwashed, crushed glass cullet properties for two sources used in their Pennsylvania study (Table 86). The moisture content of around 5.5% was a function of the debris content, which was up to about 5.5%. The median particle sizes from these sources ranged from 2.2 to 3 mm, with most of the particles between the 4.75 and 0.075 mm sieves; there was a maximum of 5% of minus 0.057 mm. The chemical composition of glass cullet will depend somewhat on the color of the glass (Table 87). The main com- pound in glass cullet, regardless of color, was silica oxide. This high silica content was one of the main concerns when using glass cullet as aggregate substitutes in PCC because of potential ASR expansion. The sodium oxide compound was the primary alkali component. engineering ProPerties Acceptable highway application physical properties can be obtained when the cullet is post-processed by washing and crushing. Research by the Florida DOT (Cosentino et al. 1995 a, b) showed that post-processed glass cullet could be used to produce a number of standard aggregate gradations (Table 88). Physical properties such as permeability, soil classifications, maximum dry density, and porosity depended on the final gradation of the gullet. The resulting gradations could be classified as either A-1-a or A-1-b by AASHTO stan- dards and by SP or SW (poorly graded or well-graded fine to coarse sand) by Unified Soil Classification System. The ability of the glass cullet aggregate to carry a load was very low, regardless of the gradation [i.e., CBR and limestone bearing ratio (LBR) values below 5]. The specific gravity also became much more consistent (2.40 to 2.55) when the cullet was washed, regardless of gradation. This led to slightly lower unit weights for the glass cullet when compared with those for typical soils (100 to 110 lb/ft3). Wartman et al. (2004) investigated the engineering prop- erties for the two Pennsylvania sources of unwashed, crushed chapter five Waste glass ByProducts

67 sources of glass cullet (Table 89). These materials met the soil classification requirements for an SW soil. Toughness is higher than for typical aggregates but below most state upper limit specifications of 30% to 40%. The laboratory evalua- tion for freeze/thaw resistance (sodium sulfate soundness) was consistent and low for up to 120 cycles. The constant head hydraulic conductivity at 90% modified Proctor density showed results similar to those of typical SW soils indicat- ing the glass cullet should be relatively free draining. Direct shear testing gave high angles of internal friction; the authors noted dilatancy behavior that increased with increasing confining stresses. A cohesion value, C, from the testing was assumed to be 0, but the triaxial testing showed that a small value was obtained. This was attributed to contamination by “gummy” substances such as labels on the glass cullet and an angle of internal friction of around 45°. Property Value AGI Visual Method (MnDOT), % 35 LOI, % 3.35 Moisture Content, % 0.1 to 2.8 Permeability, cm/s 0.41 Compacted Density, lbs/ft3 87 Specific Gravity 1.96 to 2.41 Difference in BOD, mg/l (glass compared with sand) 26 Average difference in COD, mg/l (glass compared with sand) 58 Di-n-butyl phthalate, µg/l 9.6 Krivit (1999). AGI = American Geophysical Institute; BOD = biological oxygen demand; COD = chemical oxygen demand. TABLe 85 MIxeD GLASS CULLeT PROPeRTIeS (NOT CRUSHeD) TABLe 86 exAMPLe OF UNWASHeD, CRUSHeD GLASS CULLeT PROPeRTIeS FROM A PeNNSyLVANIA STUDy By WARTMAN eT AL. (2004) Test As-Received Source 1 Source 2 Ave. Range Avg. Range 23.5–94.3 22.4 06.2–30.2 63.2 % ,tnetnoC retaW 14.3–26.0 28.1 57.0–0.0 43.0 % ,tnetnoC sirbeD 84.2 ytivarG cificepS 2.49 Minimum density, lb/ft3 1.18–8.67 2.97 4.27–1.17 8.17 Gradation Information Maximum density, lb/ft3 111.7 110.7–112.3 108.6 108.5–109.2 Median grain size, D50, mm 2.24 1.85–2.62 3 2.70–3.30 Coefficient of uniformity 6.2 4.3–10.0 7.2 5.4–7.0 Sand content, (0.075 to 4.75 mm), % 91.3 89.5–93.0 70 66.5–74.0 Fines content (<0.075 mm), % 3.2 0.0–5.0 1.2 0.2–2.0 Sieve Analysis Sieve Size, mm 10 100 100 07 79 57.4 62 05 0.2 51 32 6.8 8 41 34.0 4 01 52.0 2 8 31.0 1 5 570.0 TABLe 87 CHeMICAL COMPOSITION OF GLASS CULLeT By COLOR Compound Chemical Composition, % (Oliveira et al. 2008) Chemical Composition, % (Park et al. 2004) Flint glass Amber glass Green glass Flint glass Amber glass Green glass Na2O 9.94 10.37 10.54 — — — MgO 0.75 0.81 1.18 — — — Al2O3 2.57 3.00 2.54 2.18 1.74 1.81 SiO2 74.07 73.27 72.25 71.3 72.1 73.04 Cl2O — — — — — — K2O 1.14 1.10 1.15 — — — CaO 11.53 11.36 12.35 — — — TiO2 — — — — — — Fe2O3 — — — 0.596 0.310 0.040 SO3 — — — 0.053 0.130 0.220 Cr2O3 — — — 0.44 0.01 – Na2O + K2O — — — 13.07 14.11 13.94 CaO + MgO — — — 12.18 11.52 10.75 Flint = colorless glass.

68 the trace metals in leachate (Table 91). No trace organics for waste glass were included in this report. Wartman et al. (2004) reported that their results from the TCLP or SPLP testing had values significantly lower than those for identifying hazardous wastes (Table 92). The only compounds that exceeded drinking water standards were barium, cadmium, selenium, and silver. Some variability was noted between the two sources that were attributed to miscellaneous waste stream differences such as glass color, chemical content of label ink, specialty glass chemistries, and waste thermometers (i.e., mercury content). environmentally related ProPerties Krivit (1999) reported environmental testing for 13 metals that were analyzed with no detectable amounts released from the glass samples. Ninety-two semi-volatile compounds were analyzed and only di-n-butyl pthalate had any trace amounts. The high biological oxygen demand concentrations resulted in MnDOT placing limitations for proximity of the cullet to water sources. The NCHRP 4-21 Report (Chesner et al. 2000) summarized the trace metal concentrations in waste glass (Table 90) and TABLe 88 exAMPLe OF GLASS CULLeT AFTeR POST-PROCeSSING TO MeeT ASTM D448 GRADATIONS Properties Post-Processed No. 8 No. 9 No. 10 No. 89 3.9 3.41 3.2 3.2 ytimrofinU fo tneiciffeoC 4.1 8.0 2.1 3.1 noitadarG fo tneiciffeoC D10 5.0 570.0 5.1 3 Soil Classification USCS GP SP SP SW AASHTO A-1-a A-1-a A-1-b A-1-a Modified Proctor Density, lb/ft3 Minimum 85 85 60 83 Maximum 102 105 87 111 Max. Dry Density, lb/ft3 111 18 69 16.59 78.0 300.0 45.3 64.6 s/mc ,k ,ytilibaemreP fo tneiciffeoC Specific Yield, ne 02 8 0.43 41.53 % , 1.0< 16 10.0 10.0< ruoh ,noitarutaS %58 ot emiT Direct Shear Unit weight, lb/ft3 88–93 91–101 95–108 91–103 Angle of internal friction, o 45–51 37–45 34–46 40–45 Bearing Capacity Unit weight, lb/ft3 89–95 83–97 84–98 88–107 CBR 0.9–2.7 0.8–2.8 0.8–1.7 0.4–3.3 LBR 1.1–3.4 1.0–3.5 0.6–2.1 0.5–4.0 Cosentino et al. (1995 a,b). CBR = California bearing ratio; LBR = limestone bearing ratio (Florida); D10 = particle size associated with 10% passing that size. TABLe 89 exAMPLe OF eNGINeeRING PROPeRTIeS FOR GLASS CULLeT (UNWASHeD, CRUSHeD) AS RePORTeD By WARTMAN eT AL. (2004) Test As-Received Source 1 Source 2 Ave. Range Avg. Range WS WS — WS SCSU ,snoitacifissalC slioS — 52 — 42 % ,noisarbA AL — 1.7 — 83.6 selcyc 021 ,ssendnuoS etafluS muidoS 01 x 16.1 s/mc ,ytivitcudnoC ciluardyH -4 — 6.45 x 10-4 — Modified Proctor Maximum dry unit weight, lb/ft3 116.9 — 111.4 Optimum moisture content, % 9.7 — 11.2 — Standard Proctor Maximum dry unit weight, lb/ft3 106.9 — 105.7 Optimum moisture content, % 12.8 — 13.6 — Direct Shear Internal Friction at Various Normal Stresses, o 0–60 — 61–63 59–62 — 60–120 — 58–61 55–59 — 120–200 — 63–68 47–55 — Consolidated Drained Triaxial Internal Friction, o — 48 47 —

69 usage and Production of Waste glass Chesner et al. (2000) reported on the production of waste glass in all states. States that were producing more than 500,000 tons per year were California, Connecticut, Florida, Illinois, New york, Ohio, Pennsylvania, and Texas. States with the lowest production of waste glass in 2000 were Alaska, Hawaii, Idaho, Maine, Montana, North Dakota, New Mexico, New Hampshire, Nevada, South Dakota, Utah, Vermont, and Virginia. All of the other states were producing between 100,000 and 500,000 tons a year. The Northeast Resource Recovery Association (2009) annual reported that a total of 10,862 tons of mixed glass was recycled. The NCHRP 4-21 report (Chesner et al. 2000) noted only limited use of glass byproducts in HMA applications by Canada and Great Britain. As of 2000, only Sweden was using fine glass byproducts in PCC applications, and only on a limited basis. agency survey results Glass cullet was commonly used by agencies in HMA and embankment applications (Table 93). At least one agency used glass cullet in all of the other seven applications included in the survey. Table 94 shows that only six states reported using or having used waste glass in more than one highway application. Fifteen states use this byproduct with a single application. Figure 17 shows the geographical distribution of the states TABLe 90 TRACe MeTAL LeACHATe CONCeNTRATIONS Chesner et al. (2000). Constituent TCLP (mg/L) SPLP (mg/L) Ag <0.1 — As <0.1 — Ba 0.27 — Cd <0.002 0.004 Cr <0.01 0.0023 Cu <0.056 0.011 Hg <0.004 0.0008 Pb <0.005 0.0092 Se — — TABLe 91 TRACe MeTALS Chesner et al. (2000). Metal Concentration (mg/kg) Cr 3.1 Cu 1.7 Fe 597 Hg <0.15 K 272 Mg 2,700 Mn 5.4 Mo 0.94 Na 439 Ni <0.77 Pb 2.9 Se <0.77 V 16.8 Zn 244 TABLe 92 TOxICITy CHARACTeRISTIC (TCLP) AND SyNTHeTIC PReCIPITATION (SPLP) LeACHING PROCeDURe ReSULTS Metal Standards TCLP (mg/L) SPLP (mg/L) U.S. EPA drinking water standarda Hazardous waste designationb (mg/L) Source 1 Source 2 Source 1 Source 2 Ag 0.05 5 0.02 0.02 0.02 0.02 As 0.05 5 0.10 0.10 0.10 0.10 Ba 2 100 0.151 0.10 0.10 0.10 Cd 0.005 1 0.01 0.01 0.01 0.01 Cr 0.1 5 0.03 0.0772 0.03 0.03 Hg 0.002 0.2 0.0002 0.0002 0.00024 0.0002 Pb 0.015 5 0.10 0.128 0.10 0.10 Se 0.05 1 0.20 0.20 0.20 0.20 Wartman et al. (2004). Note: All data in milligrams per liter (mg/l). aU.S. EPA (1999). bSW-846, Chapter 7.4 (Revision 3, Dec. 1994).

70 indicating experience with this byproduct by the respondent filling out the survey. Only Alaska, Hawaii, and Idaho indicated some experience in the western half of the United States. literature revieW applications The original use for glass cullet was as an aggregate substitute for natural aggregates used in highway applications. Recent research has begun to focus on the use of glass powder (mostly passing the 0.075 mm sieve) as a pozzolanic replacement for cement. The information in this section documents recent national and international research. Bound—mortar xie et al. (2003) evaluated the use of glass and glass-fly ash for ASR using ASTM C 1250. At 10% glass cullet substitu- tion for natural aggregate, the ASR expansion was dependent on the size of the glass aggregate. However, the expansion of the glass-fly ash combination was not a function of glass cullet particle size. Rather, the expansion was found to be a function of the percent of the glass cullet. Glass-fly ash mortars could use a replacement of up to 100% without exceeding expansion limits. Oliveira et al. (2008) evaluated the pozzolanic reaction of finely ground glass cullet in cement mortars in Portugal. A chemistry analysis was conducted for glass, sorted by color and then ground into a fine powder. This testing showed that the glass powder satisfied the basic chemical requirements of a pozzolan but did not comply with additional requirement for alkali content (Na2O), which was high. The high alkali content was a concern when the byproduct was used in mortar and PCC applications because of the possibility of detrimental ASR expansion. The authors noted the glass cullet should be washed before grinding because previous research indicated false reactivity predictions may be obtained when testing unwashed glass cullet powders. A jaw crusher and ball mill were used to crush the glass cullet, which was then sieved into three fractions: 0.15 to 0.075 mm, 0.075 to 0.0045 mm, and less than 0.045 mm. Grinding time optimization was determined using the Blaine specific surface at the end of various grinding times (every hour for 10 h). There was a good linear correlation between an increase in specific surface with time in the ball mill with values starting at about 65 m2/kg at time 0 and increasing to about 250 m2/kg at 9 h. The finely ground glass was used as a cement replacement at 10%, 20%, 25%, 30%, and 40% of each color and size. The shape of the glass powder was char- acterized using SeM photographs. These images indicated all of the powders had an angular shape. Testing of the mortar mixes included compressive and flexural strength (eN 198-1) and ASR evaluation with mortar Manufacturing or Misc. Construction Byproducts: Is your state using, or has ever used, this byproduct in highway applications? *Waste glass: post-consumer glass byproducts Type of Glass Byproducts Used in Highway Applications Asphalt Cements or Emulsions Crack Sealants Drainage Materials Embank. Flowable Fill HMA Pavement Surface Treatment (non- structural) PCC Soil Stability Any Type 2 1 4 9 2 8 2 3 1 Embank. = embankment. TABLe 93 ReSULTS FOR AGeNCy SURVey FOR GLASS PROCeSSING ByPRODUCTS USeD IN HIGHWAy APPLICATIONS TABLe 94 STATeS USING GLASS ByPRODUCTS IN HIGHWAy APPLICATIONS IN 2009 No. of Applications States DI 9 AP 3 TV ,YN ,NM ,AM 2 1 AK, CT, FL, HI, IA, ME, NC, NH, NJ, SC, VA,WI 2009 Waste Glass 1 2 1 2 1 1 1 9 1 3 1 1 1 MA-2 CT – 1 NJ-1 NH-1 VT-2 FIGURE 17 Agencies reporting use of glass byproducts in highway applications.

71 bars (ASTM C1260). Increasing the percent of glass powder as a cement replacement resulted in a corresponding decrease in compressive strength for the 0.15 to 0.075 mm fraction. Compressive strength was similar for both the amber and flint colored glass; the green glass powder had only a slightly higher strength at 28 days. This was attributed to the slightly higher specific surface of the green glass (about 445 m2/kg) compared with the amber and flint color (about 355 m2/kg). At 90 days, the strengths were similar for all of the mixes. Pozzolanic activity was higher for the amber glass than the control mix. The activity increased with decreasing fraction size, but decreased with increased percent of glass powder, regardless of color. There was some indication the green powdered glass was slightly more reactive than the amber. The percent of expansion (ASR test) decreased with the increasing percent of powder. At 40% of either the amber or flint colored glass powders the percent expansion was similar to the control. The green glass powder had an expansion of about 0.0053% compared with about 0.0038%. Conclusions were that 30% of 0.075 to 0.045 mm fraction of powdered glass could be used as a cement replacement without any detrimental ASR effects. Bound applications—Portland cement concrete Polley (1996) studied using glass cullet and glass powder in PCC mixes. Glass cullet was used in conjunction with and without fly ash. Several properties for the glass powder and other PCC materials are compared in Table 95. The specific gravity of the glass cullet was reported as 2.15 with 0.0% water absorption. Glass powder had a slightly higher pozzolanic content than either of the fly ashes, but a significantly higher alkali content (i.e., Na2O). Table 96 shows the various combinations of glass aggregates and glass powders used in this study. The coarse glass gradation included a range of particles between the 12.5 mm (100% passing) and 90% retained above the 1.18 mm. The finer glass gradations usually had 100% passing the 1.18 mm and 90% retained on 0.75 mm. The pozzolanic con- tent (ASTM C618) was reported as about 73%, and from about 5% to 16% Na2O (alkali) for the crushed glass. The slump was dependent on the percent of crushed glass aggregate in the PCC. As the percent of glass cullet increased, the water to cementitious material ratio needed to increase to maintain a consistent slump. An increase in the percent of powdered glass, holding the percent of crushed glass con- stant, showed a decrease in slump with the increasing percent of glass powder. Crushed glass greater than about 3 mm in size and visibly identifiable as crushed glass required heavy gloves to handle the mix safely. The surfaces of the particles were difficult to get coated with the paste. Commercially crushed glass was typically less than 1.5 mm in size and resembled sub-angular sand rather than crushed glass and produced mixes much easier to handle. During field trials, mixes with only the finer glass gradations were identifiable as “workable and finishable.” In general, glass aggregates decreased workability and the coarse gradation had more loss of workability than fine glass gradations. There was a corresponding increase in water demand. The author noted that the glass has little influence on the amount of air entrain- ment needed. Greater amounts of high range water reducer (HRWR) were needed to get the desired slump; the amounts were similar to those increases needed when just using fly ash. Compressive strength/failure planes for the low alkali-fine glass gradations had similar fracture patterns as the control. At 365 days of curing, these mixes showed a sharp failure plane with shearing of the coarser glass particles evident. Mixes with glass aggregate and with/without glass powder reduced the strength by about half. The glass aggregate reduced the strength improvement over time that was typically developed TABLe 95 MATeRIAL PROPeRTIeS FOR STUDy OF GLASS POWDeR IN PCC MIxeS After Polley (1996). Material ASTM Type F Fly Ash, F Powered Glass ASTM Type C Fly Ash, C Pozzolanic Content, % >70 61.2–76.4 73 50–70 64.5 SO3, % <5 0.7–9.4 None <5 2.4 Loss on Ignition, % <6 0.2–15.2 None <6 0.2 Na2O, % <1.5 0.8–1.7 5–16 <1.5 0.5 TABLe 96 COMBINATIONS OF COARSe GLASS AND FLy ASH MIxeS Ratios of Mix Components Strength, psi* Experimental mix Control mix Experimental mix Control mix CA/FA FA/F1 5,729 3,887 12/0 0/0 6,121 2,930 12/25 0/25 5,729 3,263 36/25 0/0 6,121 2,190 90/0 0/0 5,729 1,595 90/25 0/0 6,121 1,436 24/0 0/0 5,729 3,365 24/25 0/25 6,121 2,495 24/25 0/25 6,121 3,249 24/25 0/25 7,614 6,585 20/20 0/25 7,614 7,687 Values estimated from graphs; after Polley (1996). *Long-term strength (6 to 12 months), adjusted to 6% air voids. CA = washed, coarse glass; fine glass aggregate was a consistent 35% of the total glass aggregate in the mix; FA = fly ash; F1 = fine glass powder.

72 by fly ash. The tensile strengths of the PCC mixes with glass aggregates were similar to those for the control mixes. The freeze/thaw impact on the retained stiffness showed that the glass aggregate, either fine or coarse, resulted in increased freeze/thaw damage compared with the control, with 10% to 15% reduction in stiffness at 100 cycles compared with essentially no loss for the control. ASR testing (ASTM C1260) showed that mixes with the glass aggregate increased the expansion by at least 3 to 4 times that of the control mix at 14 days, with the maximum expansion occurring for mixes with 50% glass aggregate. The reactivity had a pessimism with regard to the percent of glass in the mix. In Korea, Park et al. (2004) evaluated glass aggregate as a replacement for fine aggregate for each of three colors of glass (amber, green, and flint). The PCC mixes contained air entrainment and latex polymer admixtures. Testing included slump, air content, and compacting factor for the fresh concrete and ASR, and compressive, tensile, and flexural strength for the hardened concrete. Fresh concrete test results showed that the air contents were not significantly influenced by the type of glass, but did exhibit a linear increase in air content with an increase in the percent of glass aggregate. The slump and compacting fac- tor showed the opposite trends; that is, they decreased with increasing glass content. The authors suggested that the more angular shape of the glass aggregate was responsible for the decreased workability and increased air voids. The hardened concrete properties showed increasing relative expansion (ASR reactivity) with increasing glass content. Compressive strength decreased with glass aggre- gate contents above 30%. The styrene-butadiene-styrene (SBR) increased the compressive strength of the mixes for a given level of glass aggregate up to 10% SBR. Higher con- centrations of the SBR then resulted in a significant decrease in compressive strength. Flexural strengths generally followed the same trends as the compressive strengths. Flexural strengths were well-correlated (nonlinearly) compressive strengths using the equation: Y X X= − +0 0004 0 16222. . Where: Y = flexural strength, and X = compressive strength. Tensile strengths were well correlated (nonlinearly) with the compressive strengths using the equation: Y X X= − +0 001 0 10892. . Where: Y = flexural strength, and X = compressive strength. Shayan and xu (2006) in Australia evaluated glass powder as a pozzalanic material in concrete in field trials of slabs. Mixed glass powder (88% < 0.010 mm; surface area of 800 m2/kg) and sand-sized glass cullet were used as a cement replacement at 0%, 20%, and 30% to construct 10 slabs (Table 97). A water/ cement (w/c) ratio of 0.49 was used for all mixes. The ratios of cement:coarse aggregate:fine aggregate was 1:2.68:2.02. Both the w/c ratio and the blending ratios were held constant and the natural aggregates were substituted at various percentages of the appropriate coarse and fine glass cullet. Compressive strength target of 5,800 psi was reached by only the 20% glass power at 28 days (Table 98). However, all mixes approached a compressive strength of 7,977 psi at 404 days, despite the 30% reduction in portland cement. Drying shrinkage was below 0.075% for all mixes. Dynamic modulus (ultrasonic pulse velocities) were all above 6,000 ksi at 404 days. Some indication of reduced permeability was noted, but the authors stated that more testing was needed before any TABLe 97 exPeRIMeNTAL VARIABLeS FOR GLASS POWDeR PCC MIxeS RePORTeD By SHAyAN AND xU (2006) Concrete Description Cement SF GLP Coarse Agg. Coarse Sand Fine Sand Crushed Glass Water Mix 1 Reference mix 380 0 0 1019 576 192 0 185 Mix 2 10% SF in binder 342 38 0 1019 566 189 0 185 Mix 3 20% GLP in binder 304 0 76 1019 564 188 0 185 Mix 4 30% GLP in binder 266 0 114 1019 558 186 0 185 Mix 5 10% SF in binder; 50% CGS 342 38 0 1019 283 94 356 185 Mix 6 20% GLP in binder; 50% CGS 304 0 76 1019 282 94 355 185 Mix 7 30% GLP in binder; 40% CGS 266 0 114 1019 335 112 281 185 Mix 8 30% GLP in binder;75% CGS 266 0 114 1019 141 47 523 185 Mix 9 No GLP; 50% CGS 380 0 0 1019 288 96 363 185 Mix 10 100% cement; 50% CGS, 30% GLP by mass of cement replaced fine sand 380 0 114 1019 288 36 306 185 GLP = fine glass powder; CGS = crushed glass sand; SF = silica fume; Agg. = aggregate.

73 conclusions could be reached. No ASR expansion was noted in any of the glass powder mixes. The authors attributed this to the glass powder behaving like a pozzolan and there- fore the alkali was not available for reactivity. SeM and eDx analyses showed that the fine glass powder particles appeared to have been consumed by the paste and converted to silicon- and calcium-rich phases that also retained large amounts of sodium. Conclusions were that the alkali originally contained in the glass was bound in the paste and crystalline materials that resulted from the pozzolanic reaction of the glass powder; it was likely no longer available for ASR product for- mation. Only a limited amount of possible ettringite formation (expansive reaction) in the alkali rich areas was observed. In the United Kingdom, Taha and Nounu (2008) evaluated PCC mixes with glass powder and glass aggregate. Two sup- plemental cementitious materials, SCM, granulated ground blast furnace slag (GGBFS), and metakolin, were used as port- land cement substitutions at 60% and 10%, respectively. The glass powder used had an average particle size of 0.045 mm and a single level of glass powder (20%) was used to replace the portland cement. Mixed glass aggregates had a particle size of less than 5 mm and were implemented as-received without any post-processing. This byproduct was used at two percentages: 50% and 100% aggregate replacement. Physical material properties are shown in Table 99. Table 100 provides a comparison of the xRF chemical analysis for the materials. Glass aggregate in fresh PCC mixes showed a decrease in consistency and wet density with increasing glass aggregate percentages. The loss of consistency was attributed to the lack Concrete Description Slump, in. 28-Day Density, lb/ft3 Compressive Strength at Various Times, psi Dynamic Modulus, ksi 90 Days 220 Days 404 Days 90 Days 220 Days 404 Days Mix 1 Reference mix 3 150.3 7,324 7,687 7,687 7,397 7,687 8,122 Mix 2 10% SF in binder 3 146.1 6,019 7,252 7,252 6,019 6,092 6,237 Mix 3 20% GLP in binder 3 142.5 6,817 6,164 8,630 7,107 7,542 7,687 Mix 4 30% GLP in binder 2 145.3 5,076 5,729 5,802 5,366 6,237 6,962 Mix 5 10% SF in binder; 50% CGS 3 145.2 5,511 5,874 7,542 6,237 6,309 6,672 Mix 6 20% GLP in binder; 50% CGS 3 142.5 6,527 7,397 7,687 5,584 6,672 6,817 Mix 7 30% GLP in binder; 40% CGS 2 144.5 6,672 6,527 6,599 6,237 6,527 6,817 Mix 8 30% GLP in binder; 75% CGS 2 139.3 6,382 7,542 7,614 5,656 5,947 6,237 Mix 9 No GLP; 50% CGS 3 145.4 6,527 7,397 8,412 6,237 6,164 6,527 Mix 10 100% cement; 50% CGS, 30% GLP by mass of cement replaced fine sand 144.5 6,817 7,542 8,412 6,092 6,237 6,672 SF = silica fume; GLP = fine glass powder; CGS = crushed glass sand. TABLe 98 ReSULTS FOR GLASS POWDeR PCC MIxeS eVALUATeD By SHAyAN AND xU (2006) IN AUSTRALIA (VALUeS eSTIMATeD FROM GRAPHS IN RePORT) Material Relative Density (ton/m3) Water Absorption (%) Oven dry SSD Cementitious materials Cement 3.14 — — GGBS 2.9 — — Metakaolin 2.6 — — Glass Powder 2.51 — — Coarse aggregate (crushed limestone) 20 mm 2.66 2.67 0.6 10 mm 2.66 2.68 0.66 Fine aggregate (sand) Sea Dredge 2.6 2.63 1.0 Recycled Glass 2.5 2.51 0.6 SSD = saturated surface dry. TABLe 99 MATeRIAL PROPeRTIeS RePORTeD IN TAHA AND NOUNU STUDy (2008)

74 of fines in the glass cullet mixes. Workability was reduced because of the sharp edges and harsh texture. Segregation and bleeding were both obvious for the glass cullet mixes. Glass powder improved the workability and consistency because of the improved texture and shape of the powder particles. The wet density was still slightly reduced. The highest values of hardened properties were consistently obtained using the PCC with metakaolin SCM. All of the mixes had 28-day compressive strengths of more than 8,000 psi. PCC mixes with the GGBFS had the lowest. The glass powder improved the compressive strengths over that of the GGBFS mixers. The glass aggregate (100%) combined with the 20% glass powder mixes had slightly lower tensile strengths when compared with the control, but slightly higher values for the static modulus. In Portugal, Oliveira et al. (2008) evaluated finely ground glass as a replacement for aggregates in PCC. Reference mix was 1:0.29:1.87:3.14 of cement, fly ash, fine aggregate, and coarse aggregate with a w/c ratio of 0.6. Fine glass sand was used as a natural sand replacement at 25%, 50%, and 100% dosage rates. Testing included slump, compressive strength, expansion (ASR), and capillary sorptive, as well as water and oxygen permeability. Water and permeability testing were conducted using a permeability cell that can measure the flow of oxygen through a 5 cm diameter by 4 cm high cylindrical sample. Once the oxygen flow rate was determined, the water permeability was determined. Capillary sorptive testing was conducted by drying 7.5 × 7.5 × 15 cm bars at 140°F until the weight loss was negligible. The bars were then submerged in water and the wet weight measured at the saturated surface dry condition. Various submersion times were used (10, 20, 30, 40, 50, 60, 70, 90, 130, and 150 min). Results became linear after a few minutes of testing. The sorptivity coefficient, k, was determined using the equation: W A k t= Where: W = the amount of water adsorbed, kg; A = cross section of the specimen in contact with the water, m2; t = time in min; and k is the sorptivity coefficient of the specimen in kg/m2/ min0.5. Hardened PCC had compressive strengths that increased with age for all mixes, as expected. Mixes with amber glass sand had increasingly higher strengths with the increased percent of glass sand. Authors noted that these results were the opposite of those reported by other researchers who found glass aggregate had lower strengths than the control mix and using a high alkali content cement further decreased the strength. This was attributed to the possible influence of replacing cement with 30% fly ash in this study and changes in the amount of cement relative to the surface area of the glass aggregate. expansion of bars in NaOH solution was below the limit of 0.1%. Sorptivity decreased with age at the time of testing for all mixes. Both 25% and 50% glass sand reduced sorptivity from about 0.24 at 28 days to about 0.18. At 100% glass sand it decreased to about 0.14 at 28 days. Oxygen per- meability increased over the control for mixes with both 25% and 50% glass sand, and then decreased for the 100% glass sand mixes. At 28 days, the control oxygen permeability was about 2.25 compared with 2.50, 2.75, and 1.75, respectively. Water permeability followed a similar trend. Conclusions included a recommendation for using a HRWR to maintain adequate workability and use fly ash to mitigate the ASR expansive reaction to use glass sand as an aggregate substitute. Oxide (%) dnaS slairetaM suoititnemeC Portland cement GGBS Metakaolin Glass powder Sea dredge sand Glass cullet CaO 64.5 40.9 0.06 8.61 7.11 10.63 SiO2 20.4 35.2 55.3 72.3 78.6 72.1 Al2O3 5.63 13.2 40.9 1.04 2.55 1.78 Fe2O3 2.85 0.39 0.71 0.17 2.47 0.36 MgO 1.09 7.86 0.28 3.89 0.46 1.26 Na2O 0.18 0.29 0.15 13.31 0.42 12.4 K2O 0.64 0.5 2.08 0.52 0.64 0.64 TiO2 0.27 0.55 <0.05 <0.05 0.15 0.06 Mn2O3 0.06 0.55 <0.05 <0.05 0.1 <0.05 SrO 0.09 0.09 <0.05 <0.05 <0.05 <0.05 P2O5 0.16 <0.05 0.15 <0.05 0.09 <0.05 Cr2O3 <0.05 <0.05 <0.05 <0.05 <0.05 0.09 TABLe 100 xRF CHeMICAL ANALySIS FOR TAHA AND NOUNU STUDy (2007)

75 unbound Ho et al. (1995) reported on the research work used to develop the Florida specifications for glass byproducts in highway fill. Two sources of Florida glass cullet were evaluated in this study. Testing included a range of ASTM and Florida standards as-written including sieve analysis (ASTM D136), specific gravity (ASTM D854), direct shear (ASTM D3080), CBR (ASTM D1883), limestone bearing ratio (FDOT FM-5-515), BOD (Method 5210B five day), DO (Method 4500-OC), total phosphorous (ePA single reagent method and persulfate digestion), total dissolved solids and total suspended solids, fixed and volatile suspended solids, and total Kjeldahl nitrogen. A number of test methods were modified to specifically address glass cullet properties that differed from conven- tional aggregates. Maximum density was determined using a combination of Marshall and Proctor compaction hammers. The minimum density was determined using ASTM D4254 (vibratory compaction). Constant head permeability (ASTM D2434) was modi- fied because the test method as written maxes out the flow at 0.02 cm/s, which was constantly exceeded by the glass cullet samples. Changes included increased inflow and outflow outlets, sample diameter, and height. Four openings in the perimeter of the sample (with pinchcocks and commercial window screening) were used to replace the standard filter cloth. Leaching solid waste (ASTM D4874) was modified and was performed using 2, 4, and 6 ft columns. A 1:1 volumetric ratio of glass to water was needed to obtain good results for the highly permeable glass cullet materials. The specification recommendations at the end of this study included: • Gradation requirements with a minimum of 97% pass- ing the ½ inch sieve and a maximum of 2% passing the 0.075 mm. • Contaminates should be no more than 1% by weight of glass cullet. • Stockpile storage with time sufficient to minimize leachable materials and leachate must meet treated water standards. • Glass cullet will not be placed directly on synthetic liners, geogrids, or geotextiles or left exposed to the air for extended periods of time. • Fill will be covered with a minimum of 1 ft of topsoil or more as needed for vegetation requirements. • Glass cullet will be compacted to a minimum density of not less than 100 lb/ft3. • Standard health and safety requirements need to be met. The researchers found that glass cullet could be safely handled when it was sized to meet ASTM D448 No. 8 or finer. It was also found to be good for drainage material because of its good frictional characteristics and resistance to breakage under high confining pressures. However, the low CBR and LBR values precluded it from being recommended for use as a base or subbase course. Florida considered glass cullet as clean debris so no special permits are needed to use the byproduct as long as it was washed prior to use. The shake extraction test was used to determine organic pollutants. In 1999, the U.S. Corps of engineers (COe) produced a fact sheet that discusses the frost heave potential when using glass cullet in highway applications. Its Cold Regions Research engineering Laboratory found that in freeze/thaw testing using materials with less than 1% passing the 0.075 mm sieve had very low frost susceptibility. Adding 30% by weight to gravel did not influence wear resistance (abrasion testing) or frost susceptibility. In one case, 30% glass cullet reduced the frost susceptibility. Krivit (1999) of the Ramsey County Public Works Depart- ment in Minnesota provided a list of the six Ramsey County demonstration projects using recycled glass as an aggregate supplement, as well as project-specific comments. The author noted the bid language “. . . shall use . . .” is preferable to “. . . will be permitted . . .” to provide sufficient incentive for the contractor to use, even when there are little to no cost increases to the contractors. The downside to the required use language was that it may place an undue burden on the agency for determining the suitable availability of the glass cullet as a feedstock. The glass supply was noticeably contaminated with sod, cans, whole bottles, and miscellaneous scrap metal. This resulted in loads of recycled materials being rejected. The recycled material supplier needed to institute additional QC procedures to ensure that there was no cross contamination of waste streams. engineers, field inspection staff, and the construction contractor were generally not satisfied with the 100% use option. Reasons included: • Inadequate material control • excess moisture content • Safety concerns with larger size glass cullet particles • Difficult to compact • Offensive odor • Construction machines behaved differently on 100% glass • Possible future subsidence • Poor byproduct delivery timing. Available supply quantities were limited to approximately 8,000 tons of recycled glass per year from the one source used for the demonstration projects. Other sources might be available but very few recycling centers were considered capable of producing high-quality mixed broken glass for use as aggregate substitutes. At a 5% ratio, only 160,000 to 300,000 tons of blended aggregate could be produced per year,

76 which is substantially lower than the annual production of millions of tons of aggregates per year. Table 101 lists the pilot projects constructed with glass cullet. In Minnesota, Sibley County was paying $60 a ton for landfill glass and had a concurrent problem finding cost- effective sources of aggregates. This agency explored the use of a 10:1 ratio of gravel to glass. Both materials were fed through an aggregate crusher that produced about 100 tons of Class 5 road gravel mix. MnDOT tested the aggregate and noted the increased quality of the gravel (no test results provided) and a 1,200 ft test strip was constructed on the CSAH 6 roadway. The specification required that unwashed reclaimed glass be free draining, with a minimum depth of ground water or bedrock of 4 ft, minimum distance of 150 ft from any surface water body, and a maximum slope of 4% to any surface water body. Based on the success of pilot projects, MnDOT and the Minnesota Office of environmental Assistance developed a new specification that included the use of reclaimed glass as an option for Class 7 aggregates as a base course. Car Project Number Project #1 Project #2 Project #3 Project #4 Project #5 Project #6 Sponsor Ramsey County PWD Ramsey County PWD Ramsey County PWD Super Cycle, Inc. Ramsey County PWD City of St. Paul Month, Year 1992–1993 May–July 1997 August 1997 July 1997 June 1998 July 1998 Grade (application) Class 5 (aggregate base) Class 6 (throughout 6" aggregate base) (aggregate base) Bituminous base and Class 5 “Glasphalt” and aggregate base Select granular borrow (first 2" to 3" of subgrade) Granular borrow (first 2" of subgrade) Project Aldrich Arena parking lot Larpenteur Avenue, Phase I County Road D and Edgerton County recycling center, parking lot and tipping pad Larpenteur Avenue, Phase II Residential paving, “Thomas/ McKubin” Glass Pre- Processing Unknown Pre-crushed, screened (1/2" minus) Pre-crushed, screened (1/2" minus) Pre-crushed, screened (1/4" minus) Screened, hand- picked Screened, hand- picked Pre-Blended Unknown Yes No Yes (into bituminous and aggregate base) No No Ratio (glass to traditional aggregate) Unknown 5% 100% 5% (nominal) 100% of first subgrade lift 100% of first subgrade lift Approved By: Ramsey County Ramsey County and Mn/DOT Ramsey County only Super Cycle and Frattalone Paving Ramsey County only City of St. Paul Spec. Wording None “… may use…” “… shall use…" “… shall use…” “… shall use…" “… shall use…” Amount of Glass Used Unknown 715 tons 108 tons 10 tons est. 128 tons (net) 122 tons Performance Results Glass Supplier No change NRG (Newport RDF Facility)a No change Super Cycle No change Super Cycle No change Super Cycle Some change (some additional quality control needed) Super Cycle No change Super Cycle Aggregate Producer Unknown Carl Bolander & Sons, Inc. Super Cycle Commercial asphalt Super Cycle Super Cycle Cost Differential to Contractor Unknown $0 per ton $0 per ton Unknown $0 per ton $0 $ Paid Glass Supplier $0 $0 per ton About $1 per ton $0 About $1 per ton About $1 per ton Cost to Process Unknown Unknown Unknown Unknown Unknown Unknown Environmental Impacts Unknown None Minimal risk None Minimal risk b Min. safety riskc Krivit (1999). a “Glass” supplied by NRG, Inc. (a subsidiary of NSP) the owner–operator of the Ramsey–Washington Counties' Resource Recovery Facility located in Newport, Minnesota. bProject #5 involved a negligible increased risk of contaminated storm water run-off (see report text and Mn/DOT report for more details). cProject #6 involved a negligible increased safety risk as the result of the stockpiling of glass pile overnight by the road construction contractor in a nonsecured residential area. PWD = Public Works Department; RDF = refuse-derived fuel. TABLe 101 LIST OF PILOT PROjeCTS CONSTRUCTeD IN FLORIDA USING GLASS CULLeT

77 windshields, other car glass, light bulbs, porcelain, laboratory glass, and glass from televisions and computers were excluded. Other specification details included a requirement for crushing operations to produce a well-graded byproduct, combined gravel/glass material must meet MnDOT specification 3138, and it shall not be used as a surfacing aggregate or shoulder surface. Reclaimed glass was limited to no more than 5% of contaminates (e.g., paper, foil, metal, corks, and wood debris). Guidelines for estimating debris were also developed and based on the following normal aggregate sampling procedures. These guidelines recommended that when stockpiling whole bottles before blending, sample the glass within the glass pile before crushing (about 40 lb), after crushing (about 10 lb), before blending, and then conduct one visual inspection every 50 cubic yards of glass. The steps in the recommended visual inspection procedure were: • Select an 8 to 10 in. test pan that is about 2 in. deep. • If whole bottles are sampled, break them into 1 in. minus size so that it will fit in the test pan. • Place 1 to 3 lb of glass in the test pan and level the sample. • estimate the amount of unattached debris in the glass and compare with a reference chart for visual observations. Safeguards and material properties were controlled by asking suppliers for a letter of certification. Glass suppliers needed to meet environmental requirements and work together with counties, recyclable collection companies, and recyclable processors to ensure the QC of the byproduct. More than 15 Minnesota counties were beginning to use glass cullet in this application. In Pennsylvania, Wartman et al. (2004) determined the geotechnical properties of glass cullet and waste industrial glass. The authors defined glass cullet as post-consumer glass comprised of mixed colored glass fragments resulting from the breakage of glass containers, predominately food, juice, beer, and liquor bottles that could not be reused by bottle manufacturers. Waste industrial glass including such materials as broken, obsolete, and/or off-specification glass from the manufacturing of plate, window, and analytical glassware was also considered as part of the glass cullet definition. Glass from automobiles, lead crystal, television monitors, lighting fixtures, and electronics applications were excluded because of their composition and coatings. The properties of the glass cullet used for this research showed variable levels of contaminates depending on the source (two were evaluated), including bottle labels, metal, and plastic caps (between 0.8% and 3.4%). A summary of the results is in Table 102. The glass cullet, as-received, had variable levels of con- taminates, depending on the source, including bottle labels, metal, and plastic caps (between 0.8 and 3.4%). The glass cullet toughness was higher than for typical aggregates but below TABLe 102 MATeRIAL PROPeRTIeS FOR THe PeNNSyLVANIA PROjeCTS Test As-Received Post- Compaction As-Received Post- Compaction 2 ecruoS 1 ecruoSAve. Range Avg. Range 14.3–26.0 23.5–94.3 22.4 — 06.2–30.2 63.2 % ,tnetnoC retaW — — 28.1 — 57.0–0.0 43.0 % ,tnetnoC sirbeD — — 94.2 — — 84.2 ytivarG cificepS 57.1–27.1 03.1–32.1 72.1 — 61.1–41.1 51.1 ytisneD muminiM Gradation Information Maximum density 1.79 1.77–1.80 – 1.74 – — Median grain size, D50, mm 2.24 1.85–2.62 1.6 3 2.70–3.30 — Coefficient of uniformity 6.2 4.3–10.0 6.5 7.2 5.4–7.0 — Sand content, (0.075 to 4.75 m), % 91.3 89.5–93.0 87 70 66.5–74.0 — Fines content (< 0.075 mm), % 3.2 0.0–5.0 6.2 1.2 0.2–2.0 — Sieve Analysis Sieve size, mm 10 100 — 100 100 — 100 4.75 97 — 97 70 — 77 2 50 — 55 26 — 41 8.6 23 — 32 15 — 23 0.43 14 — 21 8 — 14 0.25 10 — 17 4 — 9 0.13 8 — 11 2 — 7 0.075 5 — 7 1 — 4 Soils Classifications USCS SW — SW SW SW Wartman et al. (2004). SW = well-graded sand.

78 most state upper limit specifications, and the freeze/thaw resis- tance was good with no more than 7.1% loss after 120 cycles (Table 103). Constant head hydraulic conductivity at 90% modified Proctor density showed results similar to those of typical SW natural soils indicating the glass cullet should be relatively free draining. Direct shear testing showed dilatancy behavior, which increased with increasing confin- ing stresses. Direct shear testing showed that a small value was obtained for cohesion that was attributed to contamination by “gummy” substances such as labels on the glass cullet. Testing considerations were needed for small punctures in the membrane. The initial membrane was greased and a second membrane was then slipped over the first. A correction fac- tor was applied to account for the additional stiffness of the two membranes. environmental testing was conducted using both the TCLP and SPLP. Some variability was noted between the two sources, which was attributed to miscellaneous waste stream differences such as glass color, chemical content of label ink, specialty glass chemistries, and waste thermometers (i.e., mercury content). The test results are shown in Table 104. Barriers noted included contracting mechanisms, existing nonperformance (material specific)-based specifications, and unnecessary cross referencing of specifications. An example was provided for one Philadelphia recycler who accepted glass and accumulated approximately 1,000 tons per month of glass TABLe 103 eNGINeeRING PROPeRTIeS FOR THe PeNNSyLVANIA PROjeCTS Tests As-Received Source 1 Source 2 Ave. Range Ave. Range — 52 — 42 % ,noisarbA AL — 1.7 — 83.6 ssendnuoS etafluS muidoS 01 x 16.1)s/mc( ytivitcudnoC ciluardyH -4 6.45 x 10-4 Modified Proctor Maximum dry unit weight, lb/ft3 116.9 — 111.4 — Optimum moisture content, % 9.7 — 11.2 — Standard Proctor Maximum dry unit weight, lb/ft3 106.9 — 105.7 — Optimum moisture content, % 12.8 — 13.6 — Direct Shear Internal Friction at Various Normal Stresses, o 0–60 — 61–63 — 59–62 60–120 — 58–61 — 55–59 120–200 — 63–68 — 47–55 Consolidated Drained Triaxial Internal Friction, o — 48 47 — Wartman et al. (2004). TABLe 104 TOxICITy CHARACTeRISTIC SyNTHeTIC PReCIPITATION LeACHING PROCeDURe ReSULTS FOR PeNNSyLVANIA STUDy Trace Metal U.S. EPA Drinking Water Standarda (mg/L) Hazardous Waste Designationb (mg/L) TCLP (mg/L) SPLP (mg/L) Source 1 Source 2 Source 1 Source 2 Arsenic 0.05 5 0.10 0.10 0.10 0.10 Barium 2 100 0.151 0.10 0.10 0.10 Cadmium 0.005 1 0.01 0.01 0.01 0.01 Chromium 0.1 5 0.03 0.0772 0.03 0.03 Lead 0.015 5 0.10 0.128 0.10 0.10 Mercury 0.002 0.2 0.0002 0.0002 0.00024 0.0002 Selenium 0.05 1 0.20 0.20 0.20 0.20 Silver 0.05 5 0.02 0.02 0.02 0.02 Wartman et al. (2004). Note: All data in milligrams per liter (mg/L). aU.S. EPA (1999). bSW-846, Chapter 7.4 (Revision 3, Dec. 1994).

79 cullet, but had difficulty in identifying reuse applications. The recycler ended up land filling the material at a cost of $18 a ton. New york City suspended glass collection in 2002 because of a lack of reuse applications. The lack of standard test methods for characterization of material properties was also a problem. These authors would rather see the use of performance specification limits. Another example was of a local municipal engineer who also served as the recycling coordinator, who could not approve of the use of crushed glass for septic field drainage because that regulation was under the control of the Department of environmental Protection, which in turn required the use of DOT approved aggregates. Barriers related to specifications included material property values specified for natural materials only; considerations of different values for recycled material were not considered. For example, minimum density requirements (e.g., embankment soil requirements) failed to account for lower specific gravities of glass byproducts. Also, some specifications placed arbitrary limits, or limits based on old research, on the amount of glass. In some cases, the total exclusion of glass in the embankment material was also a barrier. The environmental Works (2003) in Washington reported that Washington has allowed aggregate blends with up to 15% glass cullet in ballast, shoulder ballast, crushed surface base coarse, aggregate for gravel base, gravel backfill for foun- dations (classes A and B), gravel backfill for walls, gravel backfill for pipe bedding, gravel backfill for drains, backfill for sand drains, sand drainage blankets, gravel borrow, bed- ding material for rigid and flexible pipe, foundation materials (classes A, B, and C), and bank run gravel for trench back- fill. Also, 100% glass aggregate was allowed by WSDOT for backfill for walls, pipe bedding, sand drains, sand blanket, and bedding material for flexible pipe. This report noted difficulties with using glass cullet in PCC applications that were related to ASR reactions. The use of blast furnace slag was noted as a possibility for reducing this problem, which would reduce the total alkalis in the mixture. Handling concerns focused on the potential hazards associated with fugitive dust (eye contact and inhalation). Bottle glass was derived from an amorphous or noncrystalline silica, but was classified by OSHA only as a “nuisance” dust. Dampening the cullet helped mitigate the dust problems. johnson (2006) provided a presentation on the Montana recycling programs. Glass cullet was post-processed using a pulverizer that was mounted on a trailer with its own genera- tor that operates on biodiesel (Figure 18). This unit produced glass cullet both 3⁄8 and 1⁄8 inch minus byproducts. The resulting glass cullet had a rounded, rather than angular, glass particle that improved the handling safety of the byproduct. Barriers that needed to be overcome included haul dis- tance, haul costs, low tipping fees, byproduct perception, project funding, and low quantities in one location. These barriers were overcome by establishing partnerships between federal, state, county, city, non-profits, and tribal agencies. Pilot projects included a glass parking lot, septic tank drain field, landscaping material, and flooring. Parking lot com- bined a GravelPave® matrix of co-joined recycled plastic rings, placed on top of a geotextile fabric and then filled with 3⁄8 inch minus glass cullet. Skumatz and Freeman (2007) provided a summary of the uses for glass cullet in a number of applications (Table 105). eight applications were noted for road, rail, and mainte- nance work. Another two geotechnical applications were also included. Clean Washington Center (1996) noted that the historical use of glass in HMA applications had been limited to county roadways with maximum speeds of 40 mph, residential streets, and parking lots. The use of glass was also limited by the cost of collecting, sorting, sizing, and transporting the byproduct. Limitations noted in the document mentioned that glass par- ticles tended to align parallel to the road surface, which resulted in reduced skid resistance. There was more of a tendency strip (i.e., have less of a bond between the asphalt and aggregate surfaces) owing to the smooth glass surfaces. Fulton (2008) reported on the use of glass cullet as an aggregate replacement in New Zealand since 2005. Factors that were expected to provide impetus for increased use of glass cullet included: • Limited permits for aggregate production being issued • Need to reduce landfill use • More costly to dump clean fill than to buy lower grade quarry material • Growing public emphasis on sustainable practices. FIGURE 18 Portable glass crushing unit used for Montana pilot projects. http://www.astswmo.org/files/meetings/2006Annual Meeting/Montana’s%20Glass%20Aggregate%20compressed %20file.pdf.

80 The benefits were noted as the reduced cost of transporting glass to landfill or distant disposal sites, reduced use of landfill air space, reduced amount of virgin aggregate consumed, and improved environmental awareness/attitudes. The costs listed as associated with glass cullet use were the costs of curbside collection, crushing glass, and mixing with aggregate. Two methods of incorporating the glass into the aggregate were evaluated. The first fed glass into the raw feed of the parent aggregate. The amount of nonglass debris was consid- erable with the first option. The larger pieces of glass posed a safety risk during placement. The second method involved crushing the glass separately then blending during aggregate crushing. The glass was screened over the 10 mm sieve, which eliminated most of the debris and improved the visual appearance of the byproduct. The oversized particles were periodically rejected and were primarily pulverized plastic and metal. Higher wear of the crushing plant as a result of the high silica content of the glass was noted. The daily outputs were low compared with 100% crushing of rock. The final gradations and the target values needed for the specification are shown in Table 106. One possible advantage to glass cullet–aggregate blends was a reduced amount of water needed for obtaining the optimum water content, which may be beneficial in water restricted areas. Performance to date showed no difference between pavement sections with or without the blended base. TABLe 106 PROPeRTIeS OF MATeRIALS USeD IN AUSTRALIAN STUDy Fulton (2008). Gradation 0% Glass 5% Glass Specification 37.5 100 100 100 19 73 74 66–81 9.5 51 54 43–57 4.75 35 38 28–43 2.26 24 25 19–33 1.18 17 18 12–25 0.6 12 12 7–19 0.3 8 8 3–14 0.15 6 4 0–10 0.075 4 4 0–7 CBR 210 270 80 min. Clay Index 2.7 2.8 3.0 max. document assessment survey Twenty-eight documents were reviewed for the use of glass cullet in highway applications. Between 23% and 28% of the documents contained information about either the byproduct or application properties, or manufacturing processes. Thirteen percent of the researchers reported chemical properties for their studies. Only 5% or less reported data for air and water quality testing. Cost information was presented in only 7% of the documents. Figure 19 provides information on the worldwide locations recently reporting glass cullet studies. Figure 20 provides a summary of the content in the literature. TABLe 105 SUMMARy OF USeS FOR GLASS CULLeT stluseR margorP sesU Base Material Some states have set specifications for road aggregates that provides for up to 10% of reclaimed glass being blended with other aggregates as a Class 5, 6, or 7 road base materials. Road Cover for Landfill Use the pulverized glass at the landfill in two capacities: (1) Larger pieces can be used as a road base; (2) Smaller “sand” size glass can be used as a dust control device. Glasphalt Use 5%–10% reclaimed glass mix with asphalt and aggregate for road surface. Airport Runway 10% reclaimed glass aggregate used for airport runway and apron surface Salt/Sand Mix for Roads in Winter Striping Use glass as reflective material in road stripes or cross walks Pipe Bedding or Septic System Mounds Lay below pipes when installing or build septic mounds Skumatz and Freeman (2007). It is common, well proven, and safe. Research has shown that recycled glass can actually improve the quality of gravel in an aggregate mix for road base, and can be used up to 100% as a base in some cases. Aggregate contractors use the same machines to crush glass as they use to produce aggregate. Can enhance permeability of road surface, and decrease stream runoff. Costs savings depend on aggregate prices. Installed a 3,000 x 40 ft runway about eight years ago that has worked well. Used approximately 400 tons of glass in the construction of the runway. Passes inspection of state, and, over time, skid resistance has actually increased. Mix with sand, salt, or magnesium chloride to apply to the roads and sidewalks in the winter. Very successful as a non-slip application, it increases traction significantly for both roads and sidewalks. Very successful, but not using much tonnage. Some DOTs have discovered that paint adheres better to glass than other aggregates. Glass also lasts longer and retains sparkle longer than other materials. Works well and studies have found it to be technically sound. Good potential for recycled because color mix, labels, and residual sugars are not an issue. Good drainage qualities. Both uses are successful. As a dust control, it works better than water because it reflects the sun and keeps the ground from drying out as quickly. As a road base, the permeability of the glass is an advantage. Has been used for over 30 years on country roads, highways, and even airport runways. Lifetime, wear, slippage, and cracking have proven to be comparable to conventional surface materials. Reports of glass “popping” out of surface in heat, same as rocks, but more citizen complaints around glass. Satisfactory with the public in some communities, although not in all.

81 summary of glass ByProduct information list of Byproducts The byproduct categories needed for glass cullet are: • Processed glass aggregate (amber, green, flint colors) • Powdered glass. FIGURE 19 Locations of glass cullet research. http://en.wikipedia.org/wiki/File:BlankMap-World6.svg Waste Glass Research test Procedures The following ASTM test methods were identified in the glass cullet in highway application literature and the agency survey (Table 107). materials Preparation and Byproduct Quality control Information for handling and controlling the quality of the byproduct included: • Post-processing by washing and crushing can produce acceptable physical properties. – Material variability is significantly reduced. For exam- ple, the specific gravity becomes much more consistent (2.40 to 2.55) when the cullet is washed, regardless of gradation. – Contamination by “gummy” substances such as labels on the glass cullet needs to be removed. – Reclaimed glass is typically limited to no more than 5% of contaminates (e.g., paper, foil, metal, corks, and wood debris). – Contaminates are attributed to miscellaneous waste stream differences such as glass color, chemical FIGURE 20 Summary of uses for glass cullet in highway applications found in the literature. Number of Documents Waste Glass Binders Cost Design Emulsions Filler Full Depth Reclamation HMA Drainage Embankments Fill material Flowable fill Soil stabilization Aggregates Retaining walls Crack seals Chip seals Slurry Microsurface Interlayers Non-structural overlays Localized repairs Cement types Grouts Conventional concrete High performance concrete High strength concrete Cement types Mortar cements nonstructural concrete Pervious concrete Precast Pile grout Admixtures 0 2 4 6 8 10 12 14 16 Concrete Pavement Preservation Geotechnical Asphalt

82 content of label ink, specialty glass chemistries, and waste thermometers (i.e., mercury content). • Crushing operations are needed to produce a well- graded byproduct, which can be combined gravel/glass material, and the combination must meet specification requirements. transforming marginal materials Recent research started to focus on the use of glass powder (mostly passing the 0.075 mm sieve) as a pozzolanic replace- ment for portland cement. The glass powder satisfied the basic chemical requirements of a pozzolan but did not com- ply with additional requirement for alkali content (Na2O), which was high. However, the expansive reactions could be minimized by adding either fly ash or slag to the PCC mixes. handling concerns The following handling issues needed to be addressed when using glass cullet in highway applications: • Crushed glass greater than about 3 mm in size are visibly identifiable as crushed glass and require heavy gloves to handle the mix safely. – Glass cullet can be safely handled when it is sized to meet ASTM D448 No. 8 or finer. • Handling concerns focus on the potential hazards asso- ciated with fugitive dust (eye contact and inhalation). • Stockpile storage time sufficient to minimize leachable materials is needed. design adaptations Most of the design adaptations were focused on adjustments needed in the design of PCC mixes: • expansion is a function of the percentage of the glass cullet and needs to be considered in the design phase of the project. • As the percent increased, the water to cementitious material ratio needs to increase to maintain a consistent slump. • Air content increases linearly with an increase in the percent glass aggregate and may require adjustments to the mix design. However, some research indicates the glass has little influence on the amount of air entrain- ment needed. • A HRWR is needed to maintain adequate workability. Greater amounts of HRWR are needed to get the desired slump; the amounts were similar to those increases needed when just using fly ash. • Fly ash can be used to mitigate the ASR expansive reac- tion. Blast furnace slag can also help with the expansive reaction. HMA designs needed to address: • HMA mixes designed for low traffic volume and slower speed roadways. • Lower skid resistance of glass in the surface mixes when determining the most appropriate pavement layer for the mix. • Mix designs to assess the sensitivity of HMA mixes to moisture (i.e., stripping potential). Unbound designs needed to address: • Low CBR and LBR values preclude glass cullet from being recommended for use as a base or subbase course. • Decreases in the ability of the material to allow free draining when using unwashed glass cullet needs to be considered when designing embankments and fill. TABLe 107 TeST MeTHODS USeD TO eVALUATe GLASS CULLeT AND THeIR USe IN HIGHWAy APPLICATIONS eltiT dohteM tseT AASHTO M318 Standard Specification for Glass Cullet Use for Soil-Aggregate Base Course ASTM Standards C1250 Standard Test Method for Nonvolatile Content of Cold Liquid-Applied Elastomeric Waterproofing Membranes C1260 Standard Test Method for Potential Alkali Reactivity of Aggregates (Mortar-Bar Method) C136 Standard Test Method for Sieve Analysis of Fine and Coarse Aggregates C204 Standard Test Methods for Fineness of Hydraulic Cement by Air-Permeability Apparatus C618 Standard Specification for Coal Fly Ash and Raw or Calcined Natural Pozzolan for Use in Concrete D1883 Standard Test Method for CBR (California Bearing Ratio) of Laboratory-Compacted Soils D2434 Standard Test Method for Permeability of Granular Soils (Constant Head) D3080 Standard Test Method for Direct Shear Test of Soils Under Consolidated Drained Conditions D4254 Standard Test Methods for Minimum Index Density and Unit Weights of Soils and Calculation of Relative Density D448 Standard Classification for Sizes of Aggregate for Road and Bridge Construction D4874 Standard Test Method for Leaching Solid Material in a Column Apparatus D6023 Standard Test Method for Density (Unit Weight), Yield, Cement Content, and Air Content (Gravimetric) of Controlled Low-Strength Material D6103 Standard Test Method for Flow Consistency of Controlled Low Strength Material (CLSM) D854 Standard Test Method for Specific Gravity of Soil Solids by Water Pycnometer

83 • A minimum depth of ground water or bedrock of 4 ft, minimum distance of 150 ft away from any surface water body, and a maximum slope of 4% to any body of water. • Glass cullet should not be placed directly on synthetic liners, geogrids, or geotextiles, or be left exposed to the air for extended periods of time. construction concerns Construction issues were: • Workability may be reduced that will result in more time and effort needed to finish PCC surfaces. • Segregation and bleeding can be obvious problems with fresh PCC mixes. failures, causes, and lessons learned Lessons learned from field projects included: • Only the finer glass gradations were identifiable as “workable and finishable.” • The glass supply was noticeably contaminated with sod, cans, whole bottles, and miscellaneous scrap metal. This resulted in loads of recycled materials being rejected. The recycled material supplier needed to institute addi- tional QC procedures to ensure that there was no cross- contamination of waste streams. • engineers, field inspection staff, and the construction contractor were generally not satisfied with the 100% use in unbound options. Reasons included: – Material control was inadequate – excess moisture content – Glass that was not pre-crushed exhibited sharper edges and was a concern for worker safety because of potential infections from cuts, unknown hazards (e.g., medical waste), etc. It also resulted in flatter pieces that were more difficult to compact. – Offensive odor – Possible future subsidence. • Poor byproduct delivery timing stalled the construction process. • Construction machines behave differently on 100% glass Barriers Barriers noted included: • Contracting mechanisms • existing nonperformance (material specific)-based specifications • Unnecessary cross referencing of specifications • Haul distance • Haul costs • Low tipping fees • Byproduct perception • Project funding • Low quantities in one location. costs The benefits were noted as the reduced cost of transporting glass to landfill or distant disposal site, reduced use of landfill air space, reduced amount of virgin aggregate consumed, and improved environmental awareness/attitudes. The costs listed as associated with glass cullet use were the costs of curbside collection, crushing glass, and mixing with aggregate.